time when the machine s workload is lower than usu al so that economic losses are minimized But that says the Schaeffler expert is just one of many exam ples AI not only increases the availability of machines Ultimately such algorithms in the background of all processes from the supply chain to production control through to the pre selection of job applicants will enhance efficiency says Henn We ve already imple mented initial AI applications and there will be more The number of AI applications is not only growing at Schaeffler About a year ago Katja Grace from the Future of Humanity Institute at Oxford Uni versity interviewed 352 of the world s leading AI scientists to find out when smart machines will surpass us in cer tain activities Some of the results were to be expected For instance AI systems will surpass humans in the next ten years in translation by 2024 writing essays for high school 2026 and driving trucks 2027 ac cording to the experts With other tasks though Grace s survey shows that a lot more time will pass before this happens than expected In spite of Amazon and its new Go stores without cash regis ters there ll be no getting around human employees in retail before 2031 That will also be the year in which ma chines will start to be able to write bestsellers They ll only be able to become surgeons from 2053 on Humans in front thanks to versatility Starting in 2062 says the forecast AI will be better than humans in practically any individual activi ty However for one predictions made so far in ad vance always have to be taken with a grain of salt especially since the experts queried have already proven to be fallible They predicted that an AI would only be able to beat humans in the strategy game Go in 2027 whereas it actually happened while the survey was still being analyzed On the other hand the state ment that AI systems are better than humans in all ac tivities does not at all mean that there ll be a single system combining all these abilities and that will then be as intelligent as human beings While humans at that time will no longer be able to keep pace with an AI in all areas it does seem that on the whole the ability to do all things pretty well will still be adequate to se cure humans the top spot in the race between them and the machine that the singularity resulting in machines passing us in the fast lane will happen Many prominent people who feel that way are not working directly in the field of artifi cial intelligence says Toby Walsh from the University of New South Wales in Sydney one of the leading experts in AI research Most people working in AI like myself have a healthy skepticism for the idea of the singularity We know how hard it is to get even a little intelligence into a machine let alone to achieve recur sive self improvement writes Walsh in an op ed article for the U S maga zine Wired There are many techni cal reasons why the singularity might never happen he says We might simply run into some fundamental limits Walsh for example thinks that it s possible that there are limits to how smart you can be similar to the speed of light limiting acceleration Wolfgang Wahlster does not believe that machines will soon be smarter than we are either Both in hardware and software there are such massive differences between a current AI system and the human brain that AI is still an extremely far cry from the brain s capabilities and practical intelligence says the head of the German Research Center for Artificial Intelligence DFKI head quartered in Saarbrücken so nobody needs to fear the Terminator However that does not at all mean that AI technology necessarily produces worse results than a human being in all areas according to the AI expert The dominant victories achieved by AlphaGo an AI system of Google s subsidiary DeepMind demonstrated this last year The Chinese Ke Jie then the number one in the Go world ranking had no chance Plus machines are now outperforming us in chess checkers Scrabble and company as well as in other activities previously re served to humans and their cognitive abilities Algorithms are in the background of all processes They include image and speech recognition as well as automatically capturing information from texts or checking sensor data for the presence of specific pat terns Even today smart maintenance programs based on AI algorithms are already able to tell from such read ings when an industrial system needs to be serviced so that damage is prevented in the first place says Jürgen Henn who is responsible for Strategic IT at Schaeffler The machine then automatically orders the required spare parts and a service technician scheduled at a 16 Source IBM terabytes of random access memory are available to IBM s AI Watson 2 000 times more than to a standard laptop with an 8 gigabyte RAM 105 104

Vorschau Schaeffler tomorrow 03-2018 EN Seite 104
Hinweis: Dies ist eine maschinenlesbare No-Flash Ansicht.
Klicken Sie hier um zur Online-Version zu gelangen.